STM "TIP CHANGES" -
A POSSIBLE TOOL FOR TIP CHARACTERIZATION

PACS numbers: 73.40, 61.16, 68.35

Keywords

Introduction

In the present paper a possible explanation of "abnormal" STM images of Highly Oriented Pyrolythic Graphite (HOPG) is proposed. The interpretation of several images taken at our laboratory in atmospheric conditions, and at room temperature (for the experimental details cf. the subsequent section) containing tip changes, left-right asymmetry (cf. fig.1) is based on the work of Mizes et al [5].

Fig 1., 2. HOPG STM images

In this paper it is pointed out that the surface of HOPG has a threefold symmetry and although different images have been seen over the same area of graphite, the variety of images cannot be due to different surface properties of graphite. Our proposed method based on the assumption of Mizes et al [5], that an image given by a two atom tip may be considered as a superposition of two shifted images makes possible to draw conclusions concerning modifications of the tip structure.

Undesired, so called "tip changes" are frequently observed, i.e. drastic changes of the image appear during its acquisition. This may be considered as a consequence of the modified structure of the microtip i.e. modifications are caused in the tip cluster configuration or orientation. Once this hypothesis is accepted, a new method to investigate the structure of the STM tip cluster may be proposed. A great amount of work has been carried out on HOPG, its structure and its behaviour during STM investigation are well-known. From the modifications of the acquired image during imaging HOPG the modifications of the tip structure may be inferred.

Experiment

An oscillation between two metastable states of the tip cluster was observed for about 6 minutes (cf. fig 3) after which the image structure characteristic for the upper parts is found again, and no more negative images were observed.

Fig 3. Two scans recorded with 6 minutes difference. The left and right images correspond to left and right scan directions.

It should be emphasized that in several cases similar differences between the left and right images have been received, the image in fig.3 is only one example of our results (cf. fig.1 and 2).

In fig.4 (first image pair from fig.3, 50x50 A²) a tip change is shown where before the change the upper regions of the left and right image are identical and the atomic details are poorly resolved. After the tip change an improved resolution is found and the left and right images are negative to each other.

Fig 4. First scan of fig.3 with regions selected for further processing highligted.

The exact location of the tip change was determined using autocorrelation technique. The two images were separated into four parts: left upper (LU), left lower (LL), right upper (RU) and right lower (RL) ones. For each of these four images unit cell averaging was done.

The results are shown in fig.5. The unit cell averages show more clearly that LU is coincident with RU and RL is the negative of LL. This statement is also supported by the line cuts shown in fig 6. Inverted images of HOPG are frequently observed when using Pt tips. This image structure may exist for several tens of minutes. In ref [6] investigations of Pt clusters of 1-4 atoms on HOPG is reported. The authors conclude that the interaction of Pt atoms with HOPG is strong enough to allow the clusters to be imaged.

Fig 5. Processed selected regions of fig.4



Fig 6. Line cuts on fig.5.

Note, that in the images of the type of LU and RU the lattices of the maxima and of the minima have an identical symmetry to that of HOPG, but they are shifted with respect to each other. Hence the global structure does not have the symmetry of normal HOPG. Consequently a minimum has four maxima in its neighbourhood instead of only three as in a normal HOPG picture.

Theory

Examine the calculated tunnel current intensities from ref [1] for clusters W₁₀[111] and a W₉[111] scanned over HOPG. Line cuts in the "normal" directions, i.e. those joining a maximum with one of its closest neighbours ("short cut") show only a little difference. As seen in fig. 12a for short cuts the ratio of Half Width at Half Maximum (HWHM) to Maximum Height (MH) differ only slightly. Here A, and B stand for a and [beta] sites of HOPG. Due to the fact that the experimental data may contain noise, although the maxima are located once over a site A, in the other case over a site B, these two tip structures can hardly be distinguished.

Fig 12. Line cuts from the theoretical STM images of Ref.1

In fig. 12b the line cuts on the basis of the same calculations are shown in a direction joining a maximum with its neighbour from the second sphere of coordination ("long cut"). Here the difference between the two image structures is more clear. These observations led the authors to the conclusion that for evaluation of images the knowledge of the structure of long cuts is necessary, too.

Simulation

Fig 8. Pt15[110] cluster cutted from Pt (FCC) crystal structure

Let's assume the active microtip to be a Pt cluster oriented along the [110] direction with a one atom apex (cf. fig. 8). It is built up of 16 atoms. If the apex atom is removed a Pt₁₅ cluster is formed with a plane of four identical atoms at the corners of a rectangle which has the dimensions of a and a/Sqrt[2], where a=3.9158 Å is the lattice constant of Pt. This means that a/Sqrt[2]=2.7688 Å, close to the value of 2.46 Å which is the distance of the B sites in HOPG. If one uses a model considering each tip atom as an individual microtip and the global image as a result of the superposition of the four individual images, normal and abnormal HOPG images may be produced as a function of the rotation of the Pt [110] plane relative to the graphite lattice.

Fig 7. Crystal structure of HOPG

Fig 9. Scanning of the Pt15[110] cluster above HOPG plane

In order to reproduce the experimentally observed images, we used a "normal" HOPG image, i.e. similar to that calculated in ref [1] for a single atom tip. This was simulated using the superposition of three plane waves with propagation vectors at angles 120⁰ as used in ref [5,7]. It was found that the structure calculated in [1] is well reproduced by a combination [ cos x, cos x, (sin x+cos x) / Sqrt[2] ]. The four atom Pt [110] plane, and the simulated normal HOPG image were Fourier transformed and convoluted in a way shown in fig.10. The images obtained from the Fourier transforms of the products for different angles of rotations of the Pt Fourier transform with respect to the HOPG Fourier transform are shown in fig.11. These images are in good agreement with the experimental images.

Fig 10. Calculation of simulated images in Fourier space

Fig 11. Simulated STM images


Fig 13. Comparison of measured and simulated STM images

All the images were generated using the same wave combination, the only difference is the angle of rotation of the Pt plane. The LU-RU type structure is generated at a rotation angle of 46⁰, the negative image (RL) at 19⁰ and the nearly normal image (LL) at 5⁰.

Conclusion

References

2. H.-W. Fink, IBM J. Res. Develop. 30 (1986) 460

3. B. Schleicher. Th. Jung, H. Hug, H. Burtscher, Z. Physs. D.,19 (1991) 327

4. J. Garnaes, E. Kragh, K. A. Morch, A.R. Thölén, J. Vac. Sci. Technol. A, 8 (1990)

5. H. A. Mizes, Sang-il Park, W.A. Harison, Phys. Rev. B, 36 (1987) 4491

6. U. Müller, K. Sattler, J. Xhie, N. Venkatesvaran, G. Raina, Z. Phys. D, 19 (1991) 319

7. G. M. Shedd, P. E Russel, Surf. Sci. 266 (1992) 259